Topic Explorer¶

Topic Explorer is a Gaia feature that uses AI and natural language processing to analyze indexed content and generate a dynamic visual representation of key themes in your data.

What Topic Explorer Does¶

After a dataset is indexed, Topic Explorer analyzes text chunks and produces:

• Navigable themes — A word cloud of key topics, clickable for drill-down
• Dataset summary — An AI-generated overview of the dataset content
• Suggested questions — Contextual questions you can ask about the data, with a "Generate More" capability
• Theme filtering — A slider to filter themes by the percentage of text chunks they represent

Prerequisites¶

Topic Explorer must be enabled before creating datasets:

1. Navigate to Settings > Preferences in the Gaia UI
2. Turn on the Topic Explorer toggle
3. Click Save
4. Create your dataset — Topic Explorer will analyze it during indexing

For datasets with Continuous Indexing, topics refresh every 7 days automatically.

Multilingual Support¶

Topic Explorer supports themes and word clouds in multiple languages:

English, Dutch, French, German, Italian, Japanese, Korean, Spanish

API Integration¶

The Topic Explorer UI maps to the Gaia Discovery API endpoints. You can build your own Topic Explorer experience in your app:

Get the Theme Hierarchy¶

async with GaiaClient(api_key="...") as gaia:
    discovery = await gaia.get_discovery("my-dataset-id")
    for theme in discovery.get("themes", []):
        print(f"Theme: {theme['name']} ({theme['percentage']}%)")

The GET /dataset/{id}/discovery endpoint returns the full hierarchy of themes, sub-themes, and associated metadata.

Get a Theme Summary¶

curl -H "apiKey: YOUR_KEY" \
  "https://helios.cohesity.com/v2/mcm/gaia/dataset/{id}/discovery/{uuid}/summary"

Returns a text summary for a specific theme within the dataset.

Generate More Questions¶

import httpx

async with httpx.AsyncClient(
    base_url="https://helios.cohesity.com/v2/mcm/gaia",
    headers={"apiKey": api_key},
) as client:
    response = await client.post(
        f"/dataset/{dataset_id}/discovery/{theme_uuid}/generate-more-questions"
    )
    questions = response.json()
    for q in questions:
        print(f"  - {q}")

Building a Topic Explorer UI¶

You can create a custom Topic Explorer in your application:

1. Theme Tree — Use GET /dataset/{id}/discovery to render a tree or word cloud of themes
2. Theme Detail — When a user clicks a theme, fetch the summary and suggested questions
3. Question Chips — Display suggested questions as clickable chips that feed into your chat/search interface
4. Generate More — Add a button that calls the generate-more-questions endpoint

// Example React component structure
function TopicExplorer({ datasetId }: { datasetId: string }) {
  const [themes, setThemes] = useState<Theme[]>([]);
  const [selectedTheme, setSelectedTheme] = useState<Theme | null>(null);
  const [questions, setQuestions] = useState<string[]>([]);

  useEffect(() => {
    gaiaApi.getDiscovery(datasetId).then(setThemes);
  }, [datasetId]);

  const handleThemeClick = async (theme: Theme) => {
    setSelectedTheme(theme);
    const summary = await gaiaApi.getThemeSummary(datasetId, theme.uuid);
    // Display summary and questions
  };

  return (
    <div className="topic-explorer">
      <WordCloud themes={themes} onThemeClick={handleThemeClick} />
      {selectedTheme && (
        <ThemeDetail theme={selectedTheme} questions={questions} />
      )}
    </div>
  );
}

Next Steps¶

• Datasets & Indexing — How datasets work under the hood.
• Dataset Discovery — Advanced discovery API patterns.
• Querying & RAG — Use suggested questions in your RAG pipeline.